body canonicalization

dkim.go

@@ -2,11 +2,17 @@ package dkim
 
 import (
 	"bytes"
+	//"fmt"
 	"io/ioutil"
+	"regexp"
 	"strings"
 	"time"
 )
 
+const (
+	CRLF = "\r\n"
+)
+
 // sigOptions represents signing options
 type sigOptions struct {
 
@@ -128,6 +134,9 @@ func Sign(email *bytes.Reader, options sigOptions) (*bytes.Reader, error) {
 
 func canonicalize(emailReader *bytes.Reader, options sigOptions) (headers, body []byte, err error) {
 	var email []byte
+	body = []byte{}
+	rxReduceWS := regexp.MustCompile(`[ \t]+`)
+
 	email, err = ioutil.ReadAll(emailReader)
 	emailReader.Seek(0, 0)
 	if err != nil {
@@ -136,21 +145,37 @@ func canonicalize(emailReader *bytes.Reader, options sigOptions) (headers, body
 
 	parts := bytes.SplitN(email, []byte{13, 10, 13, 10}, 2)
 
+	if len(parts) != 2 {
+		return headers, body, ErrBadMailFormat
+	}
+
+	// Empty body
+	if len(parts[1]) == 0 {
+		parts[1] = []byte{13, 10}
+	}
+
 	canonicalizations := strings.Split(options.Canonicalization, "/")
 	// canonicalyze body
 	if canonicalizations[1] == "simple" {
 		body = bytes.TrimRight(parts[1], "\r\n")
 		body = append(body, []byte{13, 10}...)
 	} else {
-		for _, line := range bytes.Split(parts[1], []byte{10}) {
+		parts[1] = rxReduceWS.ReplaceAll(parts[1], []byte(" "))
-			println(line)
+		for _, line := range bytes.SplitAfter(parts[1], []byte{10}) {
+			line = bytes.TrimRight(line, " \r\n")
+			// Ignore all whitespace at the end of lines.  Implementations
+			// MUST NOT remove the CRLF at the end of the line.
+			if len(line) != 0 {
+				body = append(body, line...)
+				body = append(body, []byte{13, 10}...)
 			}
 		}
+	}
+	return
 
-	println(string(parts[0]))
+	/*println(string(parts[0]))
 	println("\r\n")
 	println(string(parts[1]))
-	println(string(body))
+	println(string(body))*/
-
 	return
 }

dkim_test.go

@@ -31,17 +31,32 @@ kS5vLkzRI84eiJrm3+IieUqIIicsO+WYxQs+JgVx5XhpPjX4SQjHtwEC2xKkWnEv
 	selector = "test"
 )
 
-var email = `Received: (qmail 28277 invoked from network); 1 May 2015 09:43:37 -0000
+var email = "Received: (qmail 28277 invoked from network); 1 May 2015 09:43:37 -0000" + CRLF +
-Received: (qmail 21323 invoked from network); 1 May 2015 09:48:39 -0000
+	"Received: (qmail 21323 invoked from network); 1 May 2015 09:48:39 -0000" + CRLF +
-MIME-Version: 1.0
+	"MIME-Version: 1.0" + CRLF +
-Date: Fri, 1 May 2015 11:48:37 +0200
+	"Date: Fri, 1 May 2015 11:48:37 +0200" + CRLF +
-Message-ID: <CADu37kTXBeNkJdXc4bSF8DbJnXmNjkLbnswK6GzG_2yn7U7P6w@tmail.io>
+	"Message-ID: <CADu37kTXBeNkJdXc4bSF8DbJnXmNjkLbnswK6GzG_2yn7U7P6w@tmail.io>" + CRLF +
-Subject: Test DKIM
+	"Subject: Test DKIM" + CRLF +
-From: =?UTF-8?Q?St=C3=A9phane_Depierrepont?= <toorop@tmail.io>
+	"From: =?UTF-8?Q?St=C3=A9phane_Depierrepont?= <toorop@tmail.io>" + CRLF +
-To: =?UTF-8?Q?St=C3=A9phane_Depierrepont?= <toorop@toorop.fr>
+	"To: =?UTF-8?Q?St=C3=A9phane_Depierrepont?= <toorop@toorop.fr>" + CRLF +
-Content-Type: text/plain; charset=UTF-8` + "\r\n\r\n" + `Hello world
+	"Content-Type: text/plain; charset=UTF-8" + CRLF + CRLF +
--- 
+	"Hello world" + CRLF +
-Toorop` + "\r\n\r\n\r\n\r\n\r\n\r\n\r\n"
+	"line with trailing space         " + CRLF +
+	"line with           space         " + CRLF +
+	"-- " + CRLF +
+	"Toorop  " + CRLF + CRLF + CRLF + CRLF + CRLF + CRLF
+
+var bodySimple = "Hello world" + CRLF +
+	"line with trailing space         " + CRLF +
+	"line with           space         " + CRLF +
+	"-- " + CRLF +
+	"Toorop  " + CRLF
+
+var bodyRelaxed = "Hello world" + CRLF +
+	"line with trailing space" + CRLF +
+	"line with space" + CRLF +
+	"--" + CRLF +
+	"Toorop" + CRLF
 
 func Test_NewSigOptions(t *testing.T) {
 	options := NewSigOptions()
@@ -107,3 +122,20 @@ func Test_Sign(t *testing.T) {
 	emailReader, err := Sign(emailReader, options)
 	assert.NoError(t, err)
 }
+
+func Test_canonicalize(t *testing.T) {
+	emailReader := bytes.NewReader([]byte(email))
+	options := NewSigOptions()
+	// simple/simple
+	options.Canonicalization = "simple/simple"
+	_, body, err := canonicalize(emailReader, options)
+	assert.NoError(t, err)
+	assert.Equal(t, []byte(bodySimple), body)
+
+	// relaxed/relaxed
+	options.Canonicalization = "relaxed/relaxed"
+	_, body, err = canonicalize(emailReader, options)
+	assert.NoError(t, err)
+	assert.Equal(t, []byte(bodyRelaxed), body)
+
+}

errors.go

@@ -21,5 +21,8 @@ var (
 	ErrSignBadCanonicalization = errors.New("bad Canonicalization parameter")
 
 	// Bad algorithm
-	ErrSignBadAlgo = errors.New("bar algorithm. Only rsa-sha1 or rsa-sha256 are permitted")
+	ErrSignBadAlgo = errors.New("bad algorithm. Only rsa-sha1 or rsa-sha256 are permitted")
+
+	// ErrBadMailFormat
+	ErrBadMailFormat = errors.New("bad mail formart")
 )